Guanidinium thiocyanate selective Ostwald ripening induced large grain for high performance perovskite solar cells

Organic-inorganic lead halide perovskite has become one of the most attractive materials for future low-cost high-efficiency solar technology. However, the polycrystalline nature of perovskite thin-film often possesses an exceptional density of defects, especially at grain boundaries (GBs) and film surface, limiting further improvement in the power conversion efficiency (PCE) of the perovskite device. Here, we report a simple method to reduce GBs and to passivate the surface of a methylammonium lead tri-iodide (MAPbI3) film by guanidinium thiocyanate (GUTS)-assisted Ostwald ripening post treatment. High-optoelectronic quality MAPbI3 film consisting of micron-sized grains were synthesized by post-treating a MAPbI3 film with GUTS/isopropanol solution (4 mg/mL, GUTS-4). Analysis of the electrochemical impedance spectra (EIS) of the solar cells showed that interfacial charge recombination resistance of the device based on a GUTS-4 post-treated MAPbI3 absorber film was increased by a factor of 1.15 to 2.6, depending on light illumination intensity, compared to the control MAPbI3 cell. This is consistent with results of the open-circuit voltage (Voc) decay and the light intensity dependent photovoltage evolution which shows device with GUTS treatment had one order longer charge carrier lifetime and was more ideal (ideality factor = = 1.25). Further characterization by Kelvin probe force microscope indicated that GUTS-4 treatment shifted the energetics of the MAPbI3 film by ~100 meV towards better energy level alignment with adjacent SnO2 electron transport layer, leading to a more favorable charge extraction process at the MAPbI3/SnO2 interface. As a result, the PCE of PSCs was enhanced from 14.59% to 16.37% and the hysteresis effect was mitigated.


Introduction
Recently, organic-inorganic lead halide perovskites have emerged as ideal materials for low-cost solution-processable, high-efficiency solar technologies [1]. Solar cells based on perovskites have shown sky-rocketing progress in power conversion efficiency (PCE) within seven years of development, from 3.8% in 2009 to over 22% in early 2016, making it the most commercially attractive solar cell technology at present [2,3]. The nature of am-bipolar charge transport of perovskite (perovskite materials can transport both electron and hole between cell terminals) [4][5][6] have led to the development of planar perovskite solar cell configurations in which perovskite functions as both the light absorber and charge transport layers. The structural simplicity of the planar configuration makes it even more economically viable and commercially attractive.
It is recognized that a delicate control of the optoelectronic quality of perovskite film is of vital importance to obtain a high-efficiency solar cell, especially for planar structure [6,7]. For example, pinholes and incomplete surface coverage can result in low light absorption and low-resistance shunting paths while the high density of undesirable defects causes a high-rate of charge recombination and low charge collection efficiency. This in turn entails low photovoltaic performance of the corresponding devices. To solve this issue, methods for preparing pinhole-free, uniform and highly compact films have been developed such as the anti-solvent one-step dripping method [7][8][9][10]. Fundamentally, the anti-solvent method involves in pouring of an anti-solvent onto a wet perovskite film to induce fast crystallization, resulting in formation of a smooth, fine-grained polycrystalline film which contains significant number of grain boundaries (GBs). Recent reports have suggested that, GBs in perovskite film are likely to function as: 1) centralized areas of electronic trap states and 2) pathways for ion migration [11][12][13]. The electronic trap states not only serve as barriers for charge transport, but they also enhance non-radiative recombination, severely reducing the charge carrier life time, and thus the overall photovoltaic device performance [14][15][16]. Possibly, the high density of GBs or electronic defects in the perovskite films prepared by typical deposition techniques, including the antisolvent-dripping method, is one of the reasons why the best performing PSCs still lags behind the theoretical Schocley-Queisser efficiency limit for a single junction solar cell [17]. In addition, ion migration at GBs is suspected to cause current-voltage hysteresis of perovskite devices [18]. As such, it is essential to reduce the number of GBs through increasing grain sizes and/or passivating the GBs of perovskite films to enable hysteresis-free high efficiency PSCs.
A variety of methods have been established to eliminate the GBs of perovskite films through the careful control of nucleation-growth rates by modification of underlying substrates [11,19] or altering the chemistry of precursor solutions [20,21]. Among them, methods based on tuning perovskite precursor composition seem to be the most successful. Yang et al. reported that methyammonium iodide (MAI) is an effective additive to tune the grain size of methyammonium lead tri-iodide (MAPbI 3 ) films from few hundred nanometer to few micrometer [20]. Pursuing a different direction, Carmona et al. stated that a moderate excess of PbI 2 in the MAPbI 3 precursor is beneficial for obtaining large homogeneous MAPbI 3 films [22]. Crystal growth retardants, such as lead(II) chloride (PbCl 2 ) have also been shown to be an advantageous precursor additive for grain size enlargement of MAPbI 3 films [23,24]. Besides lead(II) halide based additives, non-halide lead source such as lead(II) thiocyanate (Pb(SCN) 2 ) were also reported to be effective for enhancing crystal domain sizes of perovskite films [25][26][27][28][29].
Nevertheless, it is difficult to prepare high optoelectronic quality perovskite film with minimal defect densities, even with anti-solvent dripping method, and thus film post treatment is normally necessary [30][31][32][33][34]. Among these, the method based on post treatment of MAPbI 3 perovskite film using methylammonium bromide (MABr)/isopropanol was reported to lead to secondary crystallization of MAPbI 3 films through an Ostwald ripening mechanism [32]. The resulting MAPbI 3 films have enhanced grain size, crystallinity and yielded better PCE devices. The elegance of this finding is that post-treated perovskite films are pinhole-free, possessing large grain domain and low density of defects regardless of the quality of initial film. However, the MAPbI 3 films prepared by the MABr-assisted Ostwald ripening process often contain broad crystal size distributions even though the film has undergone an extreme heat treatment process (i.e. 150ºC for 10 minutes). To date, only a few Ostwald ripening-assisted precursors have been utilized for the post-treatment of perovskite films due to strict requirements for efficient coarsening in Ostwald-type ripening process [32]. Due to this, there still remains considerable room to explore the application of Ostwald ripening-assisted for improving the quality of perovskite film for higher power conversion efficiency.
Apart from grain size enlargement, approaches for passivation of GBs and interfaces/surface of perovskite film have also been investigated. Fullerene, an electron transport layer, was found to be an effective passivating agent which diffuses along GBs to passivate the defects [35][36][37]. Several groups have noted that PbI 2 is able to passivate GBs and interfaces of perovskite films [21,38,39]. These studies have shown that PbI 2 mainly occupies the space along GBs of perovskite film, forming an energy barrier that hinders leakage of both electrons and holes from perovskite film which reduces recombination.
Furthermore, guanidinium iodide (GuI) was also found to successfully suppressing defects at GBs of MAPbI 3 films (more reference) [40]. The use of (GuI) in MAPbI 3 precursor solutions has proven to extend the charge carrier lifetime by a factor of ten, and thus yielding PCE exceeding 17% for planar MAPbI 3 solar cell. The research showed that guanidinium ions do not incorporate in perovskite lattice, but instead, reside at the GBs of perovskite film, forming hydrogen bonds with under-coordinated iodine species and thus suppressing charge recombination. Although the benefits of GuI for defect mitigation in MAPbI 3 films is exceptional, the final structure of Gu-based products remain unclear.
Herein, we report a new method based on treatment of MAPbI 3 film using a guanidinium thiocyanate (GUTS) precursor solution. In this study, it was found that the GUTS-treatment effectively converts fined-grain nanometer-scale MAPbI 3 film into micron-sized MAPbI 3 perovskite film with a low density of GBs. We have also found that this GUTS-post treatment method successfully passivates the perovskite interface, which in turn significantly reduces surface charge recombination. The cooperation of these two effects enables improved performance of solar cells with less hysteresis.

Materials preparation
All materials were purchased from Sigma-Aldrich and used as received without further purification unless otherwise stated. Methylammonium lead tri-iodide (MAPbI 3 ) perovskite films were prepared based on Lewis acid-base adduct approach, details of which are described in the previous reports [9,10]. In brief, a MAPbI 3 perovskite precursor solution was prepared by residue. An electron transport layer based on SnO 2 (~40 nm) was deposited via spin-coating 0.1 M solution of tin(II) chloride (98%) in ethanol (96%) at 3000 rpm for 30 s in air by following the procedure reported previously [41]. The film was then annealed in air at 185 ºC for 1 hour before being cleaned with UV-Ozone for 20 mins and transferred to an Ar-filled glove box. A MAPbI 3 layer (~400 nm) was deposited onto the prepared SnO 2 layer at 4000 rpm for 25 s. The stability of the PSCs was tested by monitoring the efficiency of un-encapsulated PSCs which were stored in a desiccator (relative humidity ~33-35%) in dark. The performance of the device was measured every 3 days in ambient condition with relative humidity of 40-60%. perovskite films with x=0, 2, 4, 6, 8, 10, respectively. In accordance to the previous reports, [9,10] MAPbI 3 film synthesized by our one-step Lewis acid-base adduct method is pinhole-free, highly compact with crystallite sizes ranging from 100 nm to 500 nm (Fig. 1a). When MAPbI 3 films were treated with GUTS-2 precursor (2 mg/mL GUTS in isopropanol (IPA)), larger grains with domain size up to 800 nm were observed while the MAPbI 3 film remains uniform and compact (Fig. 1b). Upon increasing the concentration of GUTS to 4 mg/mL (Fig. 1c), clear evidence of grain growth is seen with most grains exceeding 1 µm. Compared to the pristine MAPbI 3 film, the lateral crystallite size of GUTS-4-MAPbI 3 film is five-folds larger, suggesting the effectiveness of this simple post processing procedure. As a general indication, the larger the perovskite film crystals, the higher the solar cell performance is expected because perovskite films with larger grains have lower numbers of GBs which act as barriers for charge transport between the cathode and anode in PCS [11,12,28]. With further increment of the GUTS concentration to 6 mg/mL, 8 mg/mL and 10 mg/mL, even larger grain sizes are observed (~2 µm in average with GUTS-10 treated film). However, pronounced contrast along GBs and surface of MAPbI 3 film are observed, which suggests formation of secondary phases. The amount of secondary phases increased as the concentration of GUTS increased, which is more clearly shown at lower magnification SEM images (Fig. S1). We conducted SEM-EDS mapping to determine the elementary composition over the area of grain interior and grain boundaries. The results show the atomic ratio of I/Pb is very similar surficial wrinkles are also found on the MAPbI 3 grains as the concentration of GUTS exceeded 6 mg/mL ( Fig. 1d-f). The appearance of these wrinkled textures is likely induced by strain due to massive MAPbI 3 grain coarsening and/or competitive grain growth of MAPbI 3 crystals and secondary phases [10], which may lead to high surface roughness. Before further characterization of optoelectronic properties of the MAPbI 3 film, we first investigated the fundamental reason for the significant morphological change in the films. With the knowledge that IPA is capable of dissolving MAPbI 3 via MAI-extracting (1) [42,43], we presume that this extraction is the initial stages of Ostwald ripening which could be the potential origin for the above morphological change.

Results and discussion
According to the Ostwald ripening process, once the GUTS/IPA solution is introduced to the surface of MAPbI 3 film, unstable small MAPbI 3 crystals with high surface energy will be partially dissolved into the IPA, giving up their mass so that large stable crystals can grow [32,44]. In this mechanism, the dissolution/de-nucleation rate of the unstable small MAPbI 3 crystals (depending on solubility capacity of solvent used) is extremely crucial because it determines the growth rate coefficient of stable large MAPbI 3 crystals [45]. In other words, the higher rate of de-nucleation/dissolution, the higher is the rate of crystal growth, and thus larger crystal size could be obtained according to the following equation: Where Z(D N , G) is the number of crystals per unit area, which is inversely proportional to the crystal size, D N is de-nucleation rate of unstable small crystals and G is growth rate of stable large crystals.
To verify our hypothesis, we examined the effect of IPA-only post treatment on the morphology of MAPbI 3 film. We found that marginal grain size enhancement was observed with the MAPbI 3 film treated with IPA ( Fig. S3) which is consistent with a recent study [32]. We infer that the extremely low solubility of MAPbI 3 in IPA leads to a low rate of de-nucleation/dissolution of unstable small MAPbI 3 crystals, and thus inefficient coarsening process. It has been reported by Yang et al., that methylammonium bromine (MABr) is able to tackle the problem as it favours the dissolution reaction (1) [32]. However, as stated, the intercalation of MABr and/or the I/Br exchange reaction could competitively take place, inhibiting the dissolution reaction, which in turn restricts effective Oswald ripening of MAPbI 3 film.
Similar to Branion, SCNanion has stronger interaction with Pb 2+ cation compared with Ianion, thus one can expect that GU-SCN/IPA may follow the same grain coarsening mechanism as does by MABr/IPA [46]. However, as shown in Fig. S4, compared to MABr/IPA treatment, a MAPbI 3 film prepared by GUTS/IPA treatment is smoother with grain sizes two-folds larger on average and a narrower grain size distribution. In addition, micron-sized MAPbI 3 grains were easily formed by using the GUTS post treatment even without heat treatment (Fig. S5), indicating that the Ostwald ripening process enabled by the GUTS/IPA treatment is very effective. More interestingly, there is no concentration window for GUTS/IPA which is required to control the size of MAPbI 3 grains. Instead, the higher of the concentration of GUTS/IPA is used, the larger grain is observed (Fig 1 and Fig. S6). The above observations lead us to conclude that besides the dissolution-assisted effect of GUTS (as similar to MABr), there must be another source that aids the grain growth. There is a possibility that there is a reaction of GUTS/IPA and MAPbI 3 as follows: CH 3 NH 3 PbI 3 GU-SCN (GuI(s) PbI 2 (s)) + (CH 3 NH 2 (g) HSCN(g) ) The unstable HSCN (gas) is probably expelled from the film rapidly [28] since we could not detect any observable signal of sulfur by X-ray photoelectron microscopy (XPS), inductively coupled plasma spectroscopy (ICP-MS) and Fourier transform infrared spectroscopy (FTIR) (not shown here). The produced PbI 2 and GuI and/or a compound formed by them ((GuI) x (PbI 2 ) y ) accumulates at the surface and GBs of MAPbI 3 film as the secondary phases according to Fig. 1e-f. It is very likely that the as-formed CH 3 NH 2 (MA) gas can dissolve MAPbI 3 crystal more efficiently than IPA, as suggested by previous reports [33,[47][48][49][50]. We therefore propose that the formed MA (gas) enhances the rate of de-nucleation and increases the solubility of unstable small MAPbI 3 crystals which in turn facilitates the grain coarsening. Compared to conventional method of Ostwald ripening, the dissolution of small-sized  We have further found that incorporating GUTS into MAPbI 3 precursor solution does slightly enlarge domain size of MAPbI 3 films as illustrated in Fig. S7a-d. However, compared to GUTS post-treated MAPbI 3 film, those with GUTS additive in perovskite precursor are two-folds smaller in terms of domain size (Fig. S7-d versus Fig. 1c). Besides, adding GUTS additives to the MAPbI 3 precursor leads to the formation of a considerable impurities on surface of MAPbI 3 film as shown in Fig. S7 (b, c, d), reducing the optoelectronic quality of the films. We therefore propose that GUTS-assisted Ostwald ripening posttreatment could be the best procedure for fabricating MAPbI 3 film with the optimal morphology.
In order to identify the effect of SCNanions and guanidinium cations on the morphology of MAPbI 3 , we prepared MAPbI 3 film from a precursor which contained 5% Pb(SCN) 2 (molar ratio) in relative to PbI 2 .
We have found that the film with 5% Pb(SCN) 2 in the MAPbI 3 precursor consisted of ~4 times larger grain sizes compared to the control MAPbI 3 film, confirming the effectiveness of SCNanion in terms of grain size enlargement. However, a large amount of secondary phases accumulated near grain boundary region of the film can be clearly seen (see Fig. S8-b). This is in accordance with the recent report in literature [28]. In a parallel experiment, we studied the effect of guanidinium cations on the grain evolution of MAPbI 3 by using different amount of guanidinium iodide (4 mg/ml and 6 mg/ml) in isopropanol solution in the post treatment of MAPbI 3 film. We have found that the grain sizes of GuItreated MAPbI 3 film are slightly larger than the control MAPbI 3 film ( Fig. S8-a, c, d). Furthermore, the film retained the uniformity and compactness of the original MAPbI 3 film. The above results indicate that both SCNanion and guanidinium cation play important roles in the grain size evolution of MAPbI 3 film." The absorption spectra of MAPbI 3 films prepared from different GUTS concentrations (Fig. S9) show characteristic absorption onset at around 770 nm regardless of GUTS concentration, indicating that the band gap of MAPbI 3 is not affected by GUTS treatment. Below 4 mg/mL GUTS in IPA, the MAPbI 3 film has slightly higher light absorption coefficient than the pristine one, suggesting an improved film crystallinity. Beyond that, the absorbance of MAPbI 3 films is reduced which can be ascribed to the formation of a substantial amount of secondary phases (as shown in Fig. 1d-f).

Fig. 3 XRD patterns of MAPbI 3 films with and without GUTS treatment.
X-ray diffraction (XRD) was used to structurally characterize the material phases in the MAPbI 3 films with and without GUTS-treatment (Fig. 3). As indexed in Fig. 3, the main peaks of all the films can be well assigned to the reflection of MAPbI 3 . It should be noted that the peak intensity ratio between (220) and (310) plane changes as concentration of GUTS increases, indicating the change in growth rate of the two crystallographic plane of MAPbI 3 crystals. This change could also be associated with the formation of secondary phases, which might alter crystal growth rate at particular direction. In addition, weak diffraction peaks at 8.12º, 9.82º, and 11.2º are observed when the concentration of GUTS for post treatment is over 6 mg/mL, which should be assigned to the observed secondary phases observed in topview SEM images (Fig. 1d-f).
To identify these unknown peaks, we collected XRD patterns of Gu-SCN (powder), GuI (powder) and "GuPbI 3 " films ( Fig. S11) because they are most likely the by-products of the reaction (3). A yellow GuPbI 3 film was prepared by spin-coating a precursor of GuI:PbI 2 :DMSO (1:1:1, molar ratio) in DMF followed by annealing at 100ºC for 2 mins. It is worth highlighting that no peak from Gu-SCN was found in the diffraction patterns of GUTS-treated MAPbI 3 , indicating that Gu-SCN was completely consumed in the reaction (3) with MAPbI 3 . In addition, the unknown peaks do not match any diffraction patterns of GuI or "GuPbI 3 " phases either, suggesting that the by-products induced by reaction (3)  Having shown that micron-sized MAPbI 3 films could be easily obtained through manipulation of the concentration of GUTS in the post treatment process, we proceed to integrate these films into solar cells.
The performance of these MAPbI 3 devices is compared and listed in Table 1.  (Table 1). These results are consistent with the observed reduction of the series resistance (R s ) by~37%, and the significant increase of shunt resistance (R sh ) by six-folds of the devices ( Table 1).
The lower R s implies the perovskite film has lower inter-particle contact resistance, higher internal film conductivity and better contacts with selective charge transport layers, while the larger R sh indicates the greater surface and GBs passivation [10,51]. In addition, slightly higher current densities are also witnessed as the GUTS precursor concentration increases, which is consistent with the UV-visible light absorption spectrum (Fig. S9 and Fig. S10). However, further increase of the GUTS content up to 8 mg/mL leads to a dramatic reduction of all photovoltaic parameters even though grain size of the MAPbI 3 film was increased (Table 1). This is probably  Since the main enhancement of device performance are V oc and FF, owing to the plunge of the series resistance and the soar of shunt resistance of the device (Table 1), we speculate that the GUTS post treatment probably enhances the electronic properties of the perovskite film towards reducing charge recombination rate. Aiming to interpret the dynamics of charge transfer and recombination in the solar cell devices and to verify this hypothesis, impedance spectroscopy (IS) measurements were conducted at open-circuit condition (~V oc bias) with various light illumination intensities [53].
The IS spectrum shows two semi-circles in the Nyquist plot. The IS spectrum can be well-fitted using the equivalent circuit shown in Fig. 5a (bottom right). In the equivalent circuit, the capacitive element, C g , at high-frequency region, represents the dielectric properties of the perovskite absorber layer while the resistive element, R 3 , is associated with the transport resistance of electron along the perovskite layer interface. C s and R 1 obtained at low-frequency region correspond to the properties of perovskite interfaces. More specifically, C s serves as ionic accumulation capacitance (in dark) or electronic accumulation capacitance (under light illumination) at electrode interfaces, while R 1 coupled with R 3 determines surface recombination resistance [53][54][55]. The extracted information from the IS fitting for both types of devices shows that the bulk capacitance, C g , is unchanged regardless of light illumination intensities but the interfacial charge accumulation capacitance, C s , increases linearly with the illumination intensity (Fig. 5b), while the resistive parameters follow the inverse trend with the illumination intensity ( Fig. 5c).
It is found that, compared to the Ref-MAPbI 3 solar cell, the interfacial capacitance, C s , of the GUTS-4-MAPbI 3 is slightly higher under light illumination, indicating a slightly higher level of electronic accumulation or higher carrier density at electrode interface. It has been reported that V oc of perovskite solar cell follows the trend with the C s because under light illumination, the C s can be proportionally associated with the density of minority carriers at perovskite interface [55]. The IS result observed in the darkness shows that both devices have similar C s , implying the same level of ionic accumulation at perovskite interface ( Fig.S12 and Table S2, Fig. 5c illustrates that the resistance, R 3 , of the GUTS-4-MAPbI 3 solar cell is slightly larger than that of the reference cell at the same V oc (Fig. 5c, upper). Nevertheless, when we use illumination as the reference, we have found that R 3 of the GUTS-4-MAPbI 3 solar cell is slightly smaller than that of the reference cell ( Figure S13). The ratio of R 3 of the MAPbI 3 cell to the GUTS-4-MAPbI 3 is in the range of film with GUTS-4 treatment has much less GBs compared to the reference MAPbI 3 film (Fig. 1a, c), we conclude that, the GBs probably serve as barriers for charge transport (in both the bulk and the perovskite interfaces). Once these barriers are eliminated, charge carrier can transport to SnO 2 more efficiently.
However, compared to the reference cell, the GUTS-4-MAPbI 3 cell has higher low frequency resistance, R 1 , (Fig. 5c, bottom). Interestingly, the surface recombination resistance which is the sum of R 1 and R 3 , R 1 + R 3 , of the GUTS-4-MAPbI 3 device is larger than the pristine sample, being consistent with the enhanced device performance (  (Fig. 5c). The ideality factor of the solar cells extracted from the plots of resistance vs voltage are calculated according to the equation [56].
Fitting the resistive impedance (Fig. 5b, c) shows that both solar cells have an ideality factor exceeding 1, and the GUTS-4-MAPbI 3 cell has lower ideality factor than the pristine device. For example, the fitted R 1 , These number close to c m indicate an accumulation capacitance as suggested by Zarazua et al [55]. By analysing the open-circuit voltage as a function of light illumination intensity (Fig. 6a), we confirm that besides direct/radiative recombination, trap-assisted recombination exists in both the GUTS-4-MAPbI 3 . The ideality factor, m, of both cells were calculated according to equation (6) [57,58]. However, compared to the reference cell, the contribution of trap-assisted recombination is smaller in the GUTS-4-MAPbI 3 cell since 21 mm   , which is consistent with the value of enhanced shunt resistance and surface charge recombination resistance of the device as shown above. We note that the ideality factors derived from the open-circuit voltage as a function of light illumination intensity results are slightly higher than those extracted from IS results. This slight discrepancy is probably due to the accumulation of ion/electronic carrier at contact under IS measurement [56].
Photovoltage decay is another effective method to investigate the recombination in perovskite solar cells. Fig. 6b illustrates the decay of V oc as a function of time upon removal of light illumination. Basically, a longer decay time of V oc is indicative of longer charge carrier life time [59,60], as can be found by the following equation [59].
Where  ir is instantaneous relaxation time and dV/dt is the decay rate of V oc . The V oc of both types of cells drops to nearly zero within 25 seconds (Fig. 6). However, the GUTS-4-MAPbI 3 cell shows a much slower decay pattern (Fig. 6b, red curve). The instantaneous relaxation time which was calculated from equation 7 is plotted in Fig. 6c. It is clearly shown that the instantaneous relaxation time of the GUTS-4-MAPbI 3 device is ~one order longer than the pristine sample which is indicative of a prolonged recombination period, in accordance with the larger surface capacitance, C s observed in IS (Fig. 5b). This result again confirms that the perovskite interface contacts have been improved with GUTS-4 post treatment. We are aware that light soaking prior to V oc decay measurement can give an electrostatic contribution to the photovoltage of PSCs as reported previously [61]. In our case, an illumination pulse with duration of ~2 s provided by a white light LED with low intensity equivalent to 0.15 sun was used. Therefore, we can rule out the electrostatic-assisted photovoltage decay effect. As suggested by Gottesman et al., that the interface of perovskite and selective contacts play a key role in the V oc decay pattern [61], we correlate this prolonged duration of voltage decay to the improvement of perovskite interfaces besides the decrease of defect in the bulk of the perovskite layer.  (Fig. 7a, c).
The mean value of contact potential differences (CPD) of the GUTS-4-MAPbI 3 film is 0.1 V higher than the reference sample, indicating a down-shifted of the electron quasi-Fermi level of the perovskite film (Fig. 7b, d). It has been reported that the conduction band of SnO 2 layer is ~170 meV below that of the MAPbI 3 perovskite film. [62] Therefore, the increase of the electron quasi-Fermi level of the GUTS-4-MAPbI 3 leads to improved energy band alignment with adjacent with SnO 2 , which is likely to facilitate the electron extraction along the MAPbI 3 /SnO 2 interface, and thus higher V oc and FF of the resultant device. The solar cells were stored in a desiccator with a relative humidity of ~33% in dark.
The stability of the reference MAPbI 3 device and the GUTS-4-MAPbI 3 based device were monitored by measuring their performance every three days (Fig. 8). Both devices show similar stability patterns, where the performance increases in the first few days due to the hole-conductivity enhancement of Spiro-OMeTAD in air. After this, the performance slowly decreases, which is correlated to the degradation of MAPbI 3 layer [4]. After 60 days storage in dry air (RH 33%), the reference cell retained 87% of PCE while the GUTS-4-MAPbI 3 cell preserved 93% of PCE (Fig. 8). The better moisture stability of the GUTS-4-MAPbI 3 cell, as compared to the reference cell could be associated with the better perovskite film quality with less defects, which reduces the probability of water penetration and degradation of perovskites [47,51].

Conclusions
We have here demonstrated the effect of GUTS-assisted Ostwald ripening post treatment on the morphology and charge recombination of MAPbI 3 film and its interfaces. It was discovered that the concentration of GUTS precursor plays a key role on crystal structure, optical properties, morphology and surface potential of perovskite film. An optimal content of GUTS precursor (GUTS-4, 4 mg/mL) induced micron-sized MAPbI 3 film, effectively passivates GBs of MAPbI 3 film from recombination and positively shifted the film surface potential as confirmed by KPFM.
Further investigation of device recombination kinetics showed that the GUTS-4-MAPbI 3 solar cells have higher interfacial charge recombination resistance and lower charge extraction resistance compared to that of the reference MAPbI 3 cell, which is in excellent agreement with the results of V oc decay studies. In addition, compared to the pristine MAPbI 3 , the GUTS-4-MAPbI 3 device had lower ideality factor, suggesting less non-radiative recombination. The resultant performance of planar PSCs was improved by 20%, from average PCE of 12.7% for the reference MAPbI 3 to 15.2% for GUTS-4-MAPbI 3 cells, mainly due to the enhancement of the V oc and FF.
In addition, devices with GUTS-4-MAPbI 3 absorber showed less-pronounced hysteresis in the J-V curve and were more inert to moisture than untreated films. This work provides new insight into the mechanism that governs the morphology and optoelectronic properties of MAPbI 3 film for high energy conversion efficiency.
Hongxia Wang is an associate professor and ARC Future Fellow at Queensland University of Technology, Australia. Her main research interest is on development of new routes for low cost solar cells and energy storage devices-work that includes perovskite solar cells, thin film solar cells using earth abundant materials and supercapacitors.

Graphical Abstract
By careful optimization of guandinium thiocyanate concentration in post treatment process, a smooth, micron-sized MAPbI 3 film which is advantageous for high photovoltaic performance perovskite solar cell is attainable.